TechRadar News

Just a few weeks before the rollout of the truly excellent Samsung Galaxy Z Fold 7, I ran into an old friend who inexplicably had a Galaxy Z Fold 4. In all my years covering Samsung's folding phone efforts, I'd rarely seen one in the wild (outside of South Korea, where they are everywhere), let alone someone I knew carrying one.

In the 43 years we've known each other, I'd never known my friend to be an early adopter, but there he was with a Samsung Galaxy Z Fold 4, unfolding it, checking work email, looking at stocks, basically using it as his primary device. I looked at it and noted how the protective covering on the flexible screen was bubbled along the crease and in other areas.

This Z Fold 4 has been abused and loved in equal measure. I was just days away from the Samsung Unpacked event, where I expected to see the rumored Samsung Galaxy Z Fold Ultra. My friend and I chuckled about the sorry state of his folding phone, and he started asking me about what he should get next. I recommended he at least wait to see what this "Fold Ultra" is all about.

Fast forward a few weeks, and I could not wait to show him the Galaxy Z Fold 7 I'd been carrying. The tech introduction would have to wait, though, because we were all gathering on the beach. As you may recall, the Z Fold 7, like most of the other most recent Folds, is rated IP48 for fresh water resistance but essentially has zero dust and sand capabilities.

Beach fumble

Samsung Galaxy Z Fold 7 compared to Z Fold 4 on the right (Image credit: Lance Ulanoff / Future)

My friend arrived on the beach, and I watched as he took out his Z Fold 4, and then gasped as it slipped from his hands and dropped onto the sand.

"Buddy, that's not sand-resistant," I told him.

Gingerly picking it up, he started frantically shaking and blowing it, "I know!" He added that he usually brings it in a Ziplock bag (I've done the same with the Z Fold 6), but he also never goes to the beach anyway – except for today, that is.

While he inspected the hinge and ports, I, in an extreme example of 'things you do not need to hear in the moment,' told him how sand could completely destroy the hinge mechanism.

Attempting a mop-up of my faux pas, I excitedly told him I had something to show him: the new Samsung Galaxy Z Fold 7.

He looked at me quizzically and chided, "Hey, you told me to wait for the Ultra."

"I know," I shrugged," This is basically that phone, they added 'Ultra' features like the 200MP sensor, but it's not called that."

I could tell that the nomenclature switch or misdirection threw him.

Cozying to the Fold 7

Comparing the Galaxy Z Fold 7 (left) to the old Galaxy Z Fold 4 (right). (Image credit: Lance Ulanoff / Future)

Later, when we were far from the beach, I collected the Z Fold 7 and showed it to him. His eyes widened. Then I handed the phone to him.

"Wow," he said as he turned the two-in-one device over in his hands. He marveled at the thinness (4.2mm) and weight, and then struggled a bit to open it. I think the magnets in this Z Fold 7, which help hold the foldable phone closed, are considerably stronger than on his Z Fold 4.

He pulled out his Z Fold 4 and opened it, looking at its puckered screen. We talked a bit about the protective covering and how the Z Fold 7 has the same thing, though I would hope it does better in the long run.

"Are the screens the same size?" he asked me, and then he placed his Z Fold 4 on top of my Z Fold 7. The size difference was obvious to him, even as I said, "No, the Z Fold 7 is an 8-inch flexible display."

I noticed that his rather thick case has an empty slot for a stylus (not sure if it was an S Pen), but I recalled that he said he initially had a case that didn't even fit his pen. While I wasn't sure of just how much my friend used a stylus (the two times I saw him with the Z Fold 4, he was without it), I shared the bad news that the Z Fold 7 lost the digitizing label and no longer supports the S Pen, though he could use a dumb capacitive stylus. He didn't say anything but nodded knowingly.

We also talked a bit about the cameras as I took a photo with the 200MM camera and then zoomed in on the photo to show him how much information that sensor can collect.

Hefting the phone like his hand was a scale, my friend smiled as I told him the Z Fold 7 is, at 215 grams, lighter than a Samsung Galaxy S25 Ultra.

Still holding the Z Fold 7, he said, "Yeah, but I have to pay $2,300, right?" I corrected him, "$2,000." He considered this, but I could tell he was already convinced that the Z Fold 7 would be his next phone.

Throughout that evening, I let other old friends hold the phone. Every single one of them was surprised at the size and weight. I don't know if they, like my friend, would give up their iPhones for it, but for a devoted Z Fold owner, there may be no better upgrade, even with the loss of S Pen support and that price.

You might also like

• The Galaxy Z Fold 7 costs too much – but this could actually be a good thing for folding phones
• A YouTuber is folding and unfolding the Galaxy Z Fold 7 200,000 times in an extreme durability test livestream
• The Samsung Galaxy Z Fold 7 might just be the perfect foldable, and sets a standard for others to follow

• Einhaus Group dwindled from 170 employees and €70 million in revenue to just eight workers
• The German phone company met its fate after months of battle
• The Bitcoin ransom was recovered but never returned

German mobile phone insurance, repair and logistics company Einhaus Group has revealed the financial extent of a 2023 ransomware attack on the company.

At its peak, Einhaus operated in over 5,000 German retail stores, partnering with major telecommunication companies like Deutsche Telekom and 1&1, generating units to €70 million in annual revenue.

Then, in 2023, a ransomware group by the name of 'Royal' infiltrated systems and encrypted critical data including contracts, billing, and communications, leaving the company battling to make up losses thereafter.

Germany's Einhaus effectively taken down by one ransomware attack

The attackers left messages via office printers warning that the company had been hacked (via WA), bringing operations to a halt with systems locked. In the months that followed, the company lost millions in revenue and operational delays, bringing total damages into the mid-seven-figure range.

The company also reportedly paid a large €200,000 ransom in Bitcoin to regain access to critical data.

German cybercrime investigators have since identified three suspects, and while public prosecutors seized the ransom-paid crypto, it never got returned to the Einhaus, which blocked the company from making a fuller recovery.

Since the ransomware attack, Einhaus has been forced to cut headcount down from a peak of around 170 to just eight, selling off property and investments to make up some losses.

Three companies under Einhaus Group, including 24logistics, have now filed for insolvency, and mobile phone repair services have been discontinued.

With cyberattacks not only becoming more common, but often more costly, Einhaus Group forms part of a growing list of companies forced to shut down as a result of ransomware attacks, including the UK's Knights of Old transport company, Stoli USA and Finland's Vastaamo.

You might also like

• Ransomware gangs are now expanding to physical threats in the real world
• Downloaded something dodgy? Here's the best malware removal
• Improve digital protection with the best firewalls

• Ascendo Immersive Audio launches new 64-inch and 100-inch subs
• "The largest subwoofers in the world for home cinema"
• Expect a six-figure price tag

If, like me, you think your hi-fi or home theater system's subwoofer is pretty hot stuff, prepare to feel very small indeed. The new subwoofer from German audio firm Ascendo Immersive Audio is probably 10 times bigger than yours, and 12 times bigger than mine.

My idea of a big subwoofer is 15 inches, like the biggest KEF Kube: my own Cambridge Audio sub is a perfectly decent eight inches, and that delivers more than enough boom to fill my room. But Ascendo's new sub is so big that it'll fill any room before you turn it on. It's a whopping 100 inches in diameter, making it over 7,800 square inches of surface area.

Who's going to buy a 100-inch sub?

The new 100 Sub is designed for a very particular kind of customer: the kind who looked at the prior model, the 80 Sub (yes. that was 80 inches), and thought "oh, that's far too small and quiet". The driver alone weighs over 570lbs (260kg).

The 100 Sub moves as much air as 40 normal 18-inch subwoofers, the firm says, and there's a slightly smaller 64-inch model too. "These two new subwoofers are the largest in the world for home cinema, and are incredibly powerful, efficient and fast," Ascendo's Geoffrey Heinzel says.

Despite the ridiculous power, these aren't designed to be total brutes. There's a shallower curve than the previous 80-inch model for improved performance, optimized quad suspension for greater movement, and "advanced" materials for noise-free operation. The cabinets have been redesigned too, with a reported boost to the low-end frequency response and improved damping to reduce resonance.

The price is firmly in the "if you have to ask, you can't afford it" territory and hasn't been officially announced, but the 80 Sub was well into six figures.

You might also like

• McIntosh's giant new subwoofer is 1000W of body-blasting bass
• Dali's mega-powerful 16-inch ported subwoofer wants to shake your home theater to bits
• Why your dream home theater isn't complete without multiple subwoofers

• 14 ex-Huawei workers have been sentenced to jail over secret theft
• Former engineers went on to work at Wi-Fi chip startup Zunpai
• Huawei has been on both sides of the IP legal fence

A Shanghai court has sentenced 14 former Huawei employees to jail for stealing chip-related business secrets, sending ripples across the industry not just in China but also globally.

The employees worked for Zunpai Communication Technology – a startup founded be engineers from HiSilicon, a Huawei unit.

The court issued jail terms of up to six years and imposed further financial penalties in a July 28 ruling which landed in Huawei's favor.

Shanghai court jails former Huawei engineers over secret stealing

After leaving Huawei in 2019, Zhang Kun, a former researcher at HiSilicon, founded Zunpai in 2021 and hired former coworkers. The startup was accused of using proprietary information by Huawei, despite the fact that the company developed Wi-Fi communication chips.

According to court documents from August 2023, the Huawei subsidiary requested that the Shanghai Intellectual Property court freeze assets under Zunpai and its subsidiaries valued at 95 million yuan.

The court's decision has not been made public online, and Huawei has not made any public remarks regarding the case. According to a South China Morning Post report, the engineers may still have grounds to challenge the decision.

More broadly, it reflect a growing commitment in China toward protecting IP, with China's Supreme People's Procuratorate stating that 21,000 people were criminally prosecuted for IP crimes in 2024 alone, including a nine-year sentence for criminal copyright infringement of Lego bricks (via National Law Review).

For Huawai, though, it represents a reversal of roles. In 2019, the Shenzhen tech giant was indicted on ten counts of stealing Western technology, attempting to steal secrets, wire fraud, and obstruction of justice.

TechRadar Pro has contacted Huawei for a comment, but we did not receive an immediate response.

You might also like

• This huge US router brand could be banned from operating in most of Europe due to patent litigation
• We've listed the best job sites and best recruitment platforms
• Check out the best productivity apps

• From mid-July 2025, there's been an uptick in malicious logins
• Researchers speculate criminals found a zero-day
• Users are advised to strengthen their cybersecurity posture

There is a chance SonicWall SSL VPN devices are carrying a zero-day vulnerability that Akira’s cybercriminals discovered, and are now using in the wild.

As of mid-July this year, cybersecurity researchers Arctic Wolf Labs observed an uptick in malicious logins, all coming through SonicWall SSL VPN instances. Since some of the endpoints were fully patched at the time of the intrusion, the researchers speculate that they might contain a zero-day flaw.

However, they haven’t ruled out the possibility that the attackers just obtained a set of active login credentials from somewhere and used them to gain access.

On the FBI's radar

In any case, organizations that suffered these malicious logins also got infected with the Akira ransomware soon after.

"A short interval was observed between initial SSL VPN account access and ransomware encryption," the researchers explained. "In contrast with legitimate VPN logins which typically originate from networks operated by broadband internet service providers, ransomware groups often use Virtual Private Server hosting for VPN authentication in compromised environments."

Until SonicWall comes forward with a patch, or at least an explanation, businesses using these VPNs are advised to enforce multi-factor authentication (MFA), delete inactive and unused firewall accounts, and make sure their passwords are fresh, strong, and unique.

Akira is a ransomware strain that first appeared in March 2023, targeting businesses across various sectors. It is known for gaining the initial foothold through compromised VPN credentials and exposed services.

The group targets both Windows and Linux systems, and is known for dismantling backups to hinder recovery. As of mid-2025, Akira has been responsible for attacks on hundreds of organizations globally, including Stanford University, Nissan Australia, and Tietoevry. The group usually directs its victims to contact them via a Tor-based website.

The FBI and CISA have issued warnings about its activity, urging organizations to implement stronger network defenses and multifactor authentication.

Via The Hacker News

You might also like

• SonicWall tells admins to patch worrying SSLVPN flaw immediately
• Take a look at our guide to the best authenticator app
• We've rounded up the best password managers

We live in a time of great flux, and, true to form, the cybersecurity industry keeps growing in complexity and scope. The AI revolution of the past 2 years has seen many enterprises scrambling to equip security leaders with the tools required to combat an increasingly borderless attack surface, not to mention growing governance and regulatory requirements demanding significant attention.

Few would envy CISOs faced with these circumstances. Still, while change can be challenging to navigate, the current security climate feels like the perfect time to embrace measures that will improve software quality and reduce risk for years to come.

I work with some of the most talented, resilient security professionals on the planet, and many of them are reinforcing their security programs to flex with the contemporary threat landscape, with their development cohorts positioned as the heart of risk reduction and vulnerability elimination.

Here is what they do differently, time and time again.

Developers have precision guidance and the right tools to prioritize security

One aspect of cybersecurity rarely discussed in depth, is the notion that code-level vulnerabilities are, at their core, a human-driven issue. They are so often perpetuated by poor coding patterns and bad habits that developers have picked up throughout their careers, and these shortcuts can have devastating consequences

Make no mistake: The blame does not lie with the development teams in any organization; it is indeed the fault of the industry as a whole, and our lack of suitable response to their upskilling need.

Bug bounties and security champion programs do go some way in creating security culture pillars within an enterprise, but this is rarely enough on its own. Every day I work with CISOs who are rising above the status quo, and they prioritize an approach that takes developers on the security journey, typically with executive buy-in for these internal programs.

Their developers thrive in an environment where Just-in-Time, relevant learning pathways are emphasized, as are tools complementary to their tech stacks. This helps to break down the significant barriers developers face in contributing meaningfully to organizational security goals, and paves the way for fair security-related KPI outcomes, as well.

They are assessed on security readiness and incentivized to improve

It is rather alarming that today, we live in a world that is essentially powered by software. The recent CrowdStrike outage proved just how easily a bug can bring critical infrastructure to its knees. Despite this, developers do not have a formal security certification or verification process that clears them to work on these vital and often precarious systems the same way an architect or mechanical engineer might.

Security leaders within organizations that are committing to a higher standard of software security resilience are taking steps not just to upskill the development cohort but routinely assess their security readiness. Perhaps a Java developer has proven themselves security-confident, but they want to be deployed on a Ruby-on-Rails project, where the skills may not necessarily translate.

A modernized security program can assess the individual, identify knowledge gaps, and pair that developer with the upskilling required to be successful, ultimately allowing them to expand their career horizons on the job, leading to higher job satisfaction and better security outcomes.

We must get to a place where data-driven insights inform rapid, high-impact company decisions; after all, the cybersecurity industry doesn’t sleep, and threat actors already have an unfair advantage over security leaders struggling with everything from the skills shortage to code monoliths that are an increasing burden within the codebase.

There is an organization-wide focus on software security and quality

One of the biggest pushes towards higher software security standards has come from CISA’s Secure-by-Design guidelines. This global movement was formed across multiple world governments, including the United States, United Kingdom, Australia, Canada, and Germany.

These guidelines promote the importance of shipping secure software from the start, and seek to establish ultimate ownership of security with software vendors, as opposed to their end-users. This is a significant break from the status quo, but, if executed well, it will assist in reducing cyber risk across the board.

The best security leaders are heeding this call, and pledging their commitment to higher software standards. For most enterprises, success will require a cultural shift that prioritizes role-based security awareness, and ongoing, hands-on support for the development cohort. However, there is no better time to get serious about uplifting internal security programs, and the sooner we do, the sooner we can point to meaningful improvements.

We've featured the best encryption software.

This article was produced as part of TechRadarPro's Expert Insights channel where we feature the best and brightest minds in the technology industry today. The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/news/submit-your-story-to-techradar-pro

• Attackers can build bigger campaigns faster with generative AI
• They're also attacking enterprise agentic AI tools
• Humans are still a major weak point

New research from CrowdStrike confirms that hackers are exploiting AI to help them deliver more aggressive attacks in less time, with the tech also democratizing lesser-skilled hackers to more advanced code.

However, besides this, they're also exploiting the same AI systems that are being used by enterprises – according to CrowdStrike, hackers are targeting the tools used to build AI agents, allowing them to gain access, steal credentials, and deploy malware.

CrowdStrike is most worried about agentic AI systems, suggesting that they've now become a "core part of the enterprise attack surface."

Attackers are honing in on enterprise AI

The security company says it observed "multiple" hackers exploiting vulnerabilities in the tools used to build AI agents, which marks a major shift from patterns of old. Until now, humans have almost always been the primary entry point into a company, but now, CrowdStrike is worried that "autonomous workflows and non-human identities [are] the next frontier of adversary exploitation."

"We’re seeing threat actors use GenAI to scale social engineering, accelerate operations, and lower the barrier to entry for hands-on-keyboard intrusions," Head of Counter Adversary Operations Adam Meyers explained.

Funklocker and SparkCat are two examples of GenAI-built malware in the real world, while DPRK-nexus Famous Chollima has also been observed using generative AI to automate its insider attack program across all phases. Scattered Spider, a group believed to consist of UK and US nationals, even managed to deploy ransomware within 24 hours of accessing systems.

"Adversaries are treating these agents like infrastructure, attacking them the same way they target SaaS platforms, cloud consoles, and privileged accounts," Meyers added.

Still, even though technologies like AI are playing an increasing role in speeding up attacks, CrowdStrike found that four in five (81%) interactive intrusions were malware-free – relying on human hands on keyboards to stay undetected.

You might also like

• We've listed the best endpoint protection software
• Consider using the best firewalls for added security
• The AI-powered future of ransomware is coming soon - here's what we need to look out for

There’s growing interest in making AI more practical, especially through techniques like retrieval-augmented generation (RAG). And it’s not just AI developers or enterprise tech teams who see the value. Anyone working with large volumes of information benefits from systems that quickly surface the right reference points.

By retrieving real, relevant information, RAG helps AI stay accurate, current, and context-aware. AWS, for example, highlights RAG as a key way to enrich GenAI with the latest research, data, and updates.

There’s a lot of potential here. But from a mainstream business perspective, even the most advanced RAG-powered systems are often disconnected from the true core of enterprise knowledge: internal documents, workflows, and operational content.

To be truly enterprise-grade, AI must connect directly to the materials that power day-to-day work—contracts, invoices, business reports, onboarding docs, customer records. In other words: your document management (DM) platform.

Analyst momentum

Industry leaders agree that robust document management is foundational for successful enterprise GenAI. Gartner recently noted that GenAI outcomes rely on “relevant, high-quality, and secure information for grounding,” all of which hinge on strong enterprise content foundations.

Bain & Company echoed the point at Nvidia’s 2025 AI Developer Conference, declaring that in every successful AI deployment, “data remains the biggest challenge and the biggest opportunity.” The message is clear: without enterprise-grade content, there is no enterprise-grade AI.

Anyone familiar with enterprise content management (ECM) won’t be surprised. While much of the GenAI conversation focuses on model selection, the real game-changer lies in the data layer—and increasingly, document management is the backbone of that layer.

Modern document management isn’t just about storing or indexing files. It’s about maintaining a live, contextualized, and navigable knowledge graph of an organization’s operational memory. Enterprises have long archived, tagged, and secured content—but today, the documents themselves are more dynamic, and the tools interpreting them are more intelligent and deeply integrated.

Modern DM lets the AI query your data

That’s because modern business documents can be structured or semi-structured or unstructured, come in multiple formats, and are scattered across diverse systems like ERP, CRM, legal systems, HR systems and email platforms. This complexity demands smarter, more connected approaches to unlock their true value.

This is exactly where AI shines, but only if your documents are accessible, integrated, and well-managed. Strong document management isn’t just a nice-to-have; it’s the foundation for successful and responsible GenAI deployment.

Techniques like RAG deliver the most value when paired with a robust document management system. In fact, RAG is at its most powerful when layered with metadata search, giving users a precise way to drill into their organization’s information space.

No large language model (LLM) is trained on your company’s unique documents, so it can’t deliver truly domain-specific answers on its own. But when you pair RAG with a modern document management platform, AI can query your internal data directly, cite the exact sources, and explain how it arrived at its conclusions. That’s something generic ChatGPT-style systems simply can’t do.

Better context through DM

That’s because RAG combines the generative power of an LLM with real, enterprise-specific data; in this case, your documents, to create a “superhuman search.” Instead of relying solely on pre-trained knowledge, RAG retrieves relevant content from your own knowledge base and injects it into the AI’s response in real time.

The result? Sharper accuracy, fewer hallucinations, and, most importantly, answers grounded in your business reality, not internet generalities. The more organized, contextualized, and accessible your enterprise content is, the more effective your RAG implementation will be. But that value only materializes if your documents are in good shape to begin with.

That’s why mature document management is essential. Rather than chasing monolithic AI platforms, leading enterprises are building modular AI pipelines—combining various AI algorithms for intelligent document understanding with document intelligence, document automation, document collaboration and of course, RAG—anchored by a strong document management foundation.

In this model, document management isn’t a back-office utility. It’s what enables the shift to this superhuman search, where any business user can ask, What are the payment terms on our top five vendor contracts from last year? and get a precise, contextualized answer in seconds.

But none of this works without good data. And in today’s enterprise, that starts with good document management.

AI isn’t wizardry—it depends on strong content foundations. It can’t fix what’s disorganized or hidden. But with the right structure in place, techniques like RAG unlock real value, turning static files into dynamic, intelligent conversations.

In today’s information-saturated world, the ability to transform enterprise content into intelligent conversation is what enables AI to deliver real strategic advantage.

We've featured the best document scanning app.

This article was produced as part of TechRadarPro's Expert Insights channel where we feature the best and brightest minds in the technology industry today. The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/news/submit-your-story-to-techradar-pro

An AI Accelerator is a deep learning or neural processor created specifically for inference and to improve the performance of an AI task. While Graphics Processing Units (GPUs) are the most common type, other specialized accelerators include Tensor Processing Units (TPUs), Data Processing Units (DPUs), Field-Programmable Gate Arrays (FPGAs), and Application-Specific Integrated Circuits (ASICs).

With so many acronyms to remember (and even more accelerator types left unnamed), we will focus on GPUs because their highly parallel architecture makes them incredibly versatile for a wide range of AI workloads.

GPU acceleration

GPUs are an in-demand commodity throughout the IT industry. Manufacturers like NVIDIA and AMD are powering the world's insatiable appetite for artificial intelligence and machine learning applications.

An AI GPU accelerator is responsible for making hundreds of thousands of calculations in parallel, and it's used in every facet of AI, large language models (LLMs), data analytics, and high-performance computing.

Such widespread adoption not only highlights their critical role in advancing modern technology but also explains why GPU demand continues to outpace supply. This leads us to a crucial question: Can your cloud hosting provider truly deliver the AI GPU hosting you need? Let's dive into what that really means with 5 things you need to know.

1: Pick the GPU your workload requires

This might sound like common sense, but it’s important to understand the GPU accelerator hardware available from your hosting provider. GPUs are not created equal; they vary massively in specification and capability.

It's important to know what VRAM does, what tensor cores are, and what an NVLink interconnect does; otherwise, it's very easy to overspec and overpay for GPU resources if you don't understand exactly what's needed.

Understand your workload: Do you want to train your own AI model, use private LLMs, or perhaps need a chatbot application? Different AI tasks require different GPU specifications.

Understand data size and IO: AI models are typically trained on huge datasets, so you need a GPU that can process data at a good rate, and you need underlying storage, ideally NVMe SSDs that can keep up to prevent bottlenecks.

Consider future scalability: Ask yourself: Will your AI project grow? You need a hosting provider that can grow with you because you may need a bigger server (more memory, faster CPU) in the near future.

Get developer feedback when choosing a framework: Does the hosting environment support your preferred framework tools out of the box? Ask your devs what they want to use; popular AI frameworks include TensorFlow, PyTorch, and JAX.

2: Choose a provider with the necessary hardware platform

GPUs are very important for AI workloads, but it's also important to consider the underlying hardware and ensure that it's fit for purpose. Network architecture and interconnects are critical for maximizing the performance of AI accelerators when hosted in the cloud.

Don't forget the importance of CPUs: CPUs are still critical for AI as they control the throughput of data to all aspects of the cloud platform. You need a provider that uses the latest CPU architectures, such as Intel Xeon or AMD EPYC, and examines the number and the speed of the CPU cores.

Go beyond high-speed networking: Fiber-optic connected networking is essential for AI platforms to function with low latency. There are 3 types of networking commonly deployed:

InfiniBand provides very low-latency and high-bandwidth communication between nodes (servers) containing GPUs. This is ideal for large-scale distributed AI clusters.

NVLink is NVIDIA's high-speed interconnect for direct GPU-to-GPU communication within a single server. It's needed for multi-GPU setups and is great for preventing bottlenecks.

High-Bandwidth Ethernet (e.g., 100GbE+) offers affordable performance for distributed AI and high-performance storage.

Latency: The ultimate goal is to choose the best hardware that will achieve the lowest latency. This is crucial because rapid GPU IO prevents bottlenecks and improves efficiency.

3: Embedded cloud AI ecosystem

One major benefit of GPU hosting is the fact that it's easy to integrate with existing services from a provider. Pick managed services that work for you. Popular integrations include hooking into cloud storage layers, picking managed security services, and backups.

If you are a business, server management options are a great way to ensure optimal performance and uptime, letting you focus on developing your AI application whilst the provider manages the underlying infrastructure, load balancing, MFA, Antivirus, Intrusion Prevention, and DDoS Protection behind the scenes.

4: Cost optimization

It's essential to keep on top of your operational expenditure, especially when using GPU accelerators. Costs can spiral if you make inefficient deployment decisions, overspecify your server requirements, or leave resources running idle around the clock. Costs vary significantly, so shop around and pick a provider that offers the hardware you need for a cost that is sustainable.

Cloud GPU hosting is the way forward (unless you can afford about $40,000 for a decent-spec GPU), do not forget about all the cooling and power requirements too. Remember to optimize your instance sizing, monitor and turn off idle resources, and take advantage of multi-instance GPU capabilities, where providers essentially slice up a GPU into smaller and very affordable partitions.

5: Support, reliability, and compliance

These factors underline a provider's ability to deliver GPU hosting that meets the needs of modern business. You may run into blockers or issues that could prevent you from releasing on schedule. Having 24x7x365 support skilled in AI/ML available for when things go wrong is vital to business continuity.

Look for strong uptime guarantees and providers that back up their claims with service credits if the unexpected happens. They should also demonstrate proven redundancy capabilities, disaster recovery, and a proactive approach to monitoring to safeguard your expensive GPU operations.

Beyond that, ensure your chosen provider meets all necessary compliance standards for your industry, whether it's GDPR, HIPAA, or SOC 2. Understand where your data will reside and confirm the provider has strong security controls like encryption and proper access management in place. Finally, always clarify the shared responsibility model so you know exactly what security aspects the provider handles versus what falls to you.

Key takeaways

Choosing the right cloud GPU hosting means looking beyond just raw power. It's about picking a provider that meets your specific requirements, delivers strong, flexible server infrastructure, and offers comprehensive tools for integration.

By optimizing costs and prioritizing critical areas like expert support, reliability, and adherence to compliance standards, you ensure your operations run efficiently and without unexpected hitches.

We've featured the best cloud storage.

This article was produced as part of TechRadarPro's Expert Insights channel where we feature the best and brightest minds in the technology industry today. The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here: https://www.techradar.com/news/submit-your-story-to-techradar-pro

Looking for a different day?

A new Quordle puzzle appears at midnight each day for your time zone – which means that some people are always playing 'today's game' while others are playing 'yesterday's'. If you're looking for Monday's puzzle instead then click here: Quordle hints and answers for Monday, August 4 (game #1288).

Quordle was one of the original Wordle alternatives and is still going strong now more than 1,100 games later. It offers a genuine challenge, though, so read on if you need some Quordle hints today – or scroll down further for the answers.

Enjoy playing word games? You can also check out my NYT Connections today and NYT Strands today pages for hints and answers for those puzzles, while Marc's Wordle today column covers the original viral word game.

SPOILER WARNING: Information about Quordle today is below, so don't read on if you don't want to know the answers.

Quordle today (game #1289) - hint #1 - VowelsHow many different vowels are in Quordle today?

• The number of different vowels in Quordle today is 4*.

* Note that by vowel we mean the five standard vowels (A, E, I, O, U), not Y (which is sometimes counted as a vowel too).

Quordle today (game #1289) - hint #2 - repeated lettersDo any of today's Quordle answers contain repeated letters?

• The number of Quordle answers containing a repeated letter today is 1.

Quordle today (game #1289) - hint #3 - uncommon lettersDo the letters Q, Z, X or J appear in Quordle today?

• No. None of Q, Z, X or J appear among today's Quordle answers.

Quordle today (game #1289) - hint #4 - starting letters (1)Do any of today's Quordle puzzles start with the same letter?

• The number of today's Quordle answers starting with the same letter is 0.

If you just want to know the answers at this stage, simply scroll down. If you're not ready yet then here's one more clue to make things a lot easier:

Quordle today (game #1289) - hint #5 - starting letters (2)What letters do today's Quordle answers start with?

• R

• S

• A

• B

Right, the answers are below, so DO NOT SCROLL ANY FURTHER IF YOU DON'T WANT TO SEE THEM.

Quordle today (game #1289) - the answers

(Image credit: Merriam-Webster)

The answers to today's Quordle, game #1289, are…

• ROAST
• SLICK
• AUDIT
• BILLY

I made a right mess of things today, only just escaping with a last gasp guess of BILLY – not that I had any other possibilities left.

I’d not heard of it before, but a billy is a kind of can or pot.

Daily Sequence today (game #1289) - the answers

(Image credit: Merriam-Webster)

The answers to today's Quordle Daily Sequence, game #1289, are…

• TAROT
• STEEP
• SLEEP
• SHORN

Quordle answers: The past 20

• Quordle #1288, Monday, 4 August: MACAW, SINCE, COLON, CHIRP
• Quordle #1287, Sunday, 3 August: MOTIF, LEERY, LOFTY, BURST
• Quordle #1286, Saturday, 2 August: WARTY, PUPAL, CLEAR, SLICE
• Quordle #1285, Friday, 1 August: ACTOR, MEALY, WIDTH, ADOBE
• Quordle #1284, Thursday, 31 July: STYLE, VALET, AGONY, ALLOY
• Quordle #1283, Wednesday, 30 July: DEBAR, ADMIN, FOLIO, USAGE
• Quordle #1282, Tuesday, 29 July: BATCH, TOPIC, MURKY, BUNCH
• Quordle #1281, Monday, 28 July: CANDY, TRYST, SHIRT, FORGO
• Quordle #1280, Sunday, 27 July: TRAWL, BALER, PIANO, MINCE
• Quordle #1279, Saturday, 26 July: MUDDY, SAINT, KINKY, POLAR
• Quordle #1278, Friday, 25 July: BONUS, RESIN, CEDAR, MADAM
• Quordle #1277, Thursday, 24 July: AGONY, VERVE, GLEAN, MINUS
• Quordle #1276, Wednesday, 23 July: OZONE, PENCE, ROOMY, WIDER
• Quordle #1275, Tuesday, 22 July: OPTIC, GIDDY, VOCAL, ADULT
• Quordle #1274, Monday, 21 July: KNEED, SNAIL, PINTO, FEAST
• Quordle #1273, Sunday, 20 July: GRACE, COUNT, EGRET, GIANT
• Quordle #1272, Saturday, 19 July: EARLY, CLICK, TRITE, SPREE
• Quordle #1271, Friday, 18 July: CINCH, FOYER, FUDGE, TAFFY
• Quordle #1270, Thursday, 17 July: CRESS, TABOO, POWER, HATER

Looking for a different day?

A new NYT Connections puzzle appears at midnight each day for your time zone – which means that some people are always playing 'today's game' while others are playing 'yesterday's'. If you're looking for Monday's puzzle instead then click here: NYT Connections hints and answers for Monday, August 4 (game #785).

Good morning! Let's play Connections, the NYT's clever word game that challenges you to group answers in various categories. It can be tough, so read on if you need Connections hints.

What should you do once you've finished? Why, play some more word games of course. I've also got daily Strands hints and answers and Quordle hints and answers articles if you need help for those too, while Marc's Wordle today page covers the original viral word game.

SPOILER WARNING: Information about NYT Connections today is below, so don't read on if you don't want to know the answers.

NYT Connections today (game #786) - today's words

(Image credit: New York Times)

Today's NYT Connections words are…

• DRESS
• SCARF
• SUBWAY
• GRATE
• PIPE
• TUBE
• CUBE
• ZIP
• HEALTH
• CARROT
• UNDERGROUND
• SECRET
• METRO
• MINCE
• COAL
• SLICE

NYT Connections today (game #786) - hint #1 - group hints

What are some clues for today's NYT Connections groups?

• YELLOW: Urban trains
• GREEN: Veg reduction 
• BLUE: The weather outside is frightful 
• PURPLE: Password 

Need more clues?

We're firmly in spoiler territory now, but read on if you want to know what the four theme answers are for today's NYT Connections puzzles…

NYT Connections today (game #786) - hint #2 - group answers

What are the answers for today's NYT Connections groups?

• YELLOW: SUBTERRANEAN TRANSIT 
• GREEN: MAKE INTO SMALLER PIECES WHILE COOKING 
• BLUE: USED TO DECORATE A SNOWMAN 
• PURPLE: CODE 

Right, the answers are below, so DO NOT SCROLL ANY FURTHER IF YOU DON'T WANT TO SEE THEM.

NYT Connections today (game #786) - the answers

(Image credit: New York Times)

The answers to today's Connections, game #786, are…

• YELLOW: SUBTERRANEAN TRANSIT METRO, SUBWAY, TUBE, UNDERGROUND
• GREEN: MAKE INTO SMALLER PIECES WHILE COOKING CUBE, GRATE, MINCE, SLICE
• BLUE: USED TO DECORATE A SNOWMAN CARROT, COAL, PIPE, SCARF
• PURPLE: CODE DRESS, HEALTH, SECRET, ZIP

• My rating: Hard
• My score: Perfect

Did you know that if you were to lay all the track of the New York SUBWAY from end to end it would reach all the way to Chicago?

I know this because today’s Connections sent me down a Wikipedia rabbit hole looking at the world’s SUBTERRANEAN TRANSIT systems.

Anyway, I digress. This was a great game for me, as I completed it in color order and it made me feel smarter than I am.

This was mainly down to getting USED TO DECORATE A SNOWMAN – giving us this group in the middle of the summer is a bit sneaky, but seeing COAL and CARROT beside each other I knew what we were looking for.

Yesterday's NYT Connections answers (Monday, August 4, game #785)

• YELLOW: ARTHROPODS APHID, BEETLE, MITE, TICK
• GREEN: TREES BEECH, CEDAR, PINE, YEW
• BLUE: [LETTER] (IS) FOR __ APPLE, COOKIE, EFFORT, VENDETTA
• PURPLE: WORDS THAT SOUNDS LIKE TWO LETTERS DECAY, EASY, GEO, ZITI

What is NYT Connections?

NYT Connections is one of several increasingly popular word games made by the New York Times. It challenges you to find groups of four items that share something in common, and each group has a different difficulty level: green is easy, yellow a little harder, blue often quite tough and purple usually very difficult.

On the plus side, you don't technically need to solve the final one, as you'll be able to answer that one by a process of elimination. What's more, you can make up to four mistakes, which gives you a little bit of breathing room.

It's a little more involved than something like Wordle, however, and there are plenty of opportunities for the game to trip you up with tricks. For instance, watch out for homophones and other word games that could disguise the answers.

It's playable for free via the NYT Games site on desktop or mobile.

Looking for a different day?

A new NYT Strands puzzle appears at midnight each day for your time zone – which means that some people are always playing 'today's game' while others are playing 'yesterday's'. If you're looking for Monday's puzzle instead then click here: NYT Strands hints and answers for Monday, August 4 (game #519).

Strands is the NYT's latest word game after the likes of Wordle, Spelling Bee and Connections – and it's great fun. It can be difficult, though, so read on for my Strands hints.

Want more word-based fun? Then check out my NYT Connections today and Quordle today pages for hints and answers for those games, and Marc's Wordle today page for the original viral word game.

SPOILER WARNING: Information about NYT Strands today is below, so don't read on if you don't want to know the answers.

NYT Strands today (game #520) - hint #1 - today's themeWhat is the theme of today's NYT Strands?

• Today's NYT Strands theme is… We have a winner!

NYT Strands today (game #520) - hint #2 - clue words

Play any of these words to unlock the in-game hints system.

• VINE 
• WARD
• WING
• TEAK
• SWEET
• GRIN

NYT Strands today (game #520) - hint #3 - spangram lettersHow many letters are in today's spangram?

• Spangram has 10 letters

NYT Strands today (game #520) - hint #4 - spangram positionWhat are two sides of the board that today's spangram touches?

First side: left, 6th row

Last side: right, 1st row

Right, the answers are below, so DO NOT SCROLL ANY FURTHER IF YOU DON'T WANT TO SEE THEM.

NYT Strands today (game #520) - the answers

(Image credit: New York Times)

The answers to today's Strands, game #520, are…

• BINGO
• GIVEAWAY
• DRAWING
• SWEEPSTAKES
• LOTTERY
• SPANGRAM: LUCKY BREAK

• My rating: Hard
• My score: 1 hint

It was possibly my LUCKY BREAK that I saw the spangram before anything else today, and immediately went in search of words associated with games of chance.

Not sure what DRAWING is in there for, but I guess numbers are drawn in lotteries and BINGO, so we’ll allow it.

Incredibly, I got SWEEPSTAKES without the S as a non-game word before I realized my error. A good job today’s search wasn’t too taxing.

Yesterday's NYT Strands answers (Monday, August 4, game #519)

• RAZOR
• CAPE
• CLIPPERS
• COMB
• TRIMMER
• AFTERSHAVE
• SPANGRAM: BARBERSHOP

What is NYT Strands?

Strands is the NYT's not-so-new-any-more word game, following Wordle and Connections. It's now a fully fledged member of the NYT's games stable that has been running for a year and which can be played on the NYT Games site on desktop or mobile.

I've got a full guide to how to play NYT Strands, complete with tips for solving it, so check that out if you're struggling to beat it each day.

If you’re not including And Just Like That… season 3 in your best HBO Max shows of 2025 list, I can hardly blame you. Since its debut in 2021, the Sex and the City (SATC) sequel has had a rocky ride on screen, frequently being lambasted on social media for how it’s changed the core personalities of Carrie (Sarah Jessica Parker), Charlotte (Kristen Davis) and Miranda (Cynthia Nixon). While Miranda left Steve (David Eigenberg) in the dust and transformed into a legal intern lesbian, Charlotte felt out of touch with her helicopter parenting of kids Rock (Alexa Swinton) and Lily (Cathy Ang).

I couldn’t even tell you who any of the middling replacements for Samantha (Kim Cattrall) are, with Cattrall herself making the career move of a lifetime by staying well away from this absolutely underwhelming car crash (well, all except for that excruciating season 2 finale cameo). The point is, none of our other stars come anywhere near her, but as much as we might want Samantha’s miraculous return in the season 3 finale to scold everyone for making such terrible television, it’s not going to happen.

So, what will happen, or perhaps the more important question is: what needs to happen? If you’ve been managing to stay awake while watching the HBO Max show, you’ll have noticed Carrie is slowly edging closer to the version of herself she used to be in Sex and the City. She’s easily the biggest stumbling block in And Just Like That… (ATLJ), and if we have to watch two more episodes until it’s over, she better have the best damn fictional U-turn TV has ever seen.

And Just Like That has been rough on everyone, but Carrie Bradshaw needs to come full circle

Spoilers for And Just Like That… season 3 episodes 9 and 10 ahead.

If you’re a seasoned SATC watcher like me, you’ll have probably thought Carrie was the worst of the quartet in the 90s too… and you’d be right. Never taking responsibility for her own actions and feelings, and instead placing the blame anywhere else, she was too frustrating to consistently warm to (even if she was dressed immaculately). Watching her make the same mistakes over again was relatable and somewhat charming for the first few seasons, but after six seasons, two standalone movies and now three seasons of a sequel series, it all wears awfully thin.

That doesn’t mean Carrie hasn’t been through it in AJLT: she’s been through the fictional death of husband Big (Ron Galotti), the real-life death of Willie Garson, who played BFF Stanford, and she’s picked back up a tired-out situationship with Aidan (John Corbett). We’ve got to have some kind of sympathy for her, right? Wrong! For the bulk of ATLJ, Carrie’s been a burden on her friends, an even bigger burden on us, and the biggest burden of all to herself. All those years of living life in the city on screen, and she doesn’t seem to have learned a single lesson from them.

However, that’s slowly started to change. Just before breaking up with Aidan (finally) and striking up a sexy new attraction to Duncan (Jonathan Clarke), we see Miranda confront Carrie about her evasive defences after Charlotte’s house party in season 3 episode 9. Miranda points out the amount of continued distrust between Carrie and Aidan, and for a split second, AJLT becomes the closest thing we’ve ever seen to a true SATC revival. Finally cutting ties with what isn’t serving her is the healthiest move Carrie has made in years, but can she sustain it?

Here’s my pitch for the AJLT season 3 ending. Miranda stays with Joy (Dolly Wells) and continues to build a healthy relationship (nothing wrong with these two, so live and let live). Harry’s (Evan Handler) prostate cancer is cleared, leaving Charlotte space to thrive as the good friend and mother we know her to be, and the others… well, they can do whatever they like as we won’t be paying attention.

But Carrie needs to dump any and all men, work on herself in the therapeutic ways she’d always dismissed in SATC, and actually start to be a good person. We can’t bow out in good conscience if she still remains a threat to New York City’s female friendships, and there will truly be no evolution in her character if she leaves in an even worse state than when she started. In the most dramatic case, the best thing for Carrie would be to get out of New York completely, which could mean moving back to England with Duncan. Frankly, I don’t care where in the world she ends up, both Carrie and AJLT need to not be insufferable the first time ever in its final moments. I know it’s hard, but pretty please?

You might also like

• The DCU's most unlikely TV show is reportedly moving forward at HBO Max as Booster Gold finds its showrunner
• HBO’s Harry Potter TV show is supposed to be a fresh start, but Nick Frost’s Hagrid says otherwise
• Superman's most surprising cameo has got me asking 3 big questions about James Gunn's next DCU project

• Ransomware attacks are rising quickly in the US
• Hackers are focusing on data exfiltration
• Oil and gas is growing increasingly popular among threat actors

The US is now the ransomware capital of the world. The majority of the attacks are targeting US organizations, and in the last year, the number of attacks has dramatically spiked. These are the conclusions echoed in the 2025 Ransomware Report, published by cybersecurity experts Zscaler ThreatLabz.

Using proprietary data, as well as samples and information collected from the wider internet, Zscaler’s researchers determined that 50% of all ransomware attacks in the last year happened in the United States, “significantly outpacing” Canada (5%) and the UK (4%).

Even when you combine all the attacks reported across the top 15 most-targeted countries, there are fewer than 3,671 that were reported in the US.

Stealing without encrypting

The number of attacks is also increasing. Year-on-year, it is up by 146% in the US, with manufacturing (1,063), technology (922), and healthcare (672) being the most-targeted industries, mostly for the potential for operational disruption, the sensitive of the stolen data, and the risk of regulatory pressure and reputational damage. Companies in the oil and gas sector saw a “staggering” increase in ransomware attacks - 900% year-on-year.

Zscaler also said that ransomware actors are increasingly abandoning the encryption part of the attack, and are focusing solely on data theft. In the last year, 10 of the biggest ransomware groups exfiltrated 238 TB of data, up 92% from last year’s 123 TB.

Right now, the biggest names in the ransomware space are RansomHub (833 victims), Akira (520), and Clop (488), but the number of threat actors is also rising. In the last year alone, the researchers identified 34 newly active ransomware families, bringing the total number up to 425.

Ransomware “flourishes” in environments with fragmented security, limited visibility, implicit trust, and outdated legacy architectures, Zscaler stresses, urging businesses to mitigate these threats by adopting a cloud-native, AI-driven, zero-trust architecture.

You might also like

• US Authorities Issue RansomHub Ransomware Alert
• Take a look at our guide to the best authenticator app
• We've rounded up the best password managers

It's almost time to return to Nevermore Academy. The second season of Wednesday, Netflix's massively popular TV take on The Addams Family franchise, is just a couple of days from making its debut. So, you'll want to know when Netflix's most-watched English language TV Original will return to a screen near you.

Below, I'll tell you when Wednesday season 2 will be released worldwide. Towards the end of this article, I'll also explain why the supernatural mystery series' next entry won't arrive in full on launch day and give you a full rundown of its complete release schedule.

So, throw on your favorite black ensemble, and get ready to get kooky and snap your fingers, and let's dive in together.

What is the release date and launch time for Wednesday season 2?

Airhorns at the ready, everyone – Wednesday Addams' TV show returns this, well, Wednesday (Image credit: Netflix)

Shocking no-one, Wednesday season 2 will be released on its weekday namesake. Indeed, one of the best Netflix shows' next semester at Nevermore will begin on Wednesday, August 6. It'll land on that date worldwide, too, so US viewers won't be able to watch it a day earlier on the world's best streaming service.

As for when new episodes of the Jenna Ortega-starring TV show will air, they'll roll out simultaneously across the world, albeit at different times. To find out when they'll drop where you live, check out the below list (NB: if your country isn't included, compare the 12am PT time slot to your time zone to determine when they'll be released where you live):

• US – 12am PT / 3am ET
• UK – 8am BST
• Australia – 5pm AEST
• New Zealand – 7pm NZST
• Canada – 12am PT / 3am ET
• India – 12:30pm IST
• Singapore – 3pm SGT
• Japan – 4pm JST
• Brazil – 4am BRT

How many episodes are there in Wednesday season 2 part 1?

Wednesday's self-anointed bestie Enid is back for more shocking events in season 2 (Image credit: Netflix)

Wednesday season 2 volume 1 will comprise four episodes. They'll all drop on August 6, too, so you won't have to wait for new chapters to arrive every week.

As for why season 2 isn't launching in full: Netflix has made a point of releasing its biggest TV hits in two or three parts. Indeed, the most recent seasons of Stranger Things, Bridgerton, Cobra Kai, Emily in Paris, The Sandman, and Outer Banks have been split into various volumes.

In fact, Squid Game is the only one of its most popular TV Originals that didn't launch in two parts (you can make the argument that seasons 2 and 3 were two halves of the same installment, though). It's not a total surprise, then, that Wednesday is getting the same treatment as many of its peers.

When will Wednesday season 2 volume 2 launch on Netflix?

We'll see much more of The Addams Family in Wednesday's latest season (Image credit: Netflix)

The wait for Wednesday season 2's final four episodes might feel like a long one, but it won't be. That's because season 2 part 2 will launch on Netflix on Wednesday, September 3.

That quartet won't be the hit show's final episodes, either. On July 23, Netflix confirmed Wednesday season 3 had been greenlit. Prepare yourself for more murder-mystery adventures with the eponymous albeit reluctant hero in the future, then.

For more on the Tim Burton-directed series' return, check out our dedicated guide on Wednesday season 2 and read the section below for further coverage.

You might also like

• Netflix reveals new trailer for Wednesday season 2 that sees Jenna Ortega's reluctant hero on a race against time to save her best friend's life
• Wednesday season 2's sneak peek teases bucketloads of Addams Family lore for the beloved Netflix series
• I didn't finish Wednesday on Netflix, but Lady Gaga’s reported cameo in season 2 is the only thing I need to give the show another shot

• Apple's CFO said data center spend could "grow substantially"
• The company posted a 10% rise in quarterly earnings
• It's also staying on top of its sustainability credentials

Apple has alluded to wanting to increase spending on artificial intelligence and data centers to catch up with rivals after early Apple Intelligence features have left users wanting more in the shadow of Microsoft and Google systems.

CEO Tim Cook expressed an interest in acquiring larger AI companies after the company bought out seven smaller companies in 2025, marking a shift from the iPhone maker's typical strategy of low-key acquisitions.

The news came as Apple announced its third-quarter financial results, revealing a 10% year-over-year quarterly revenue of $94.0 billion.

Apple plans to invest more in Apple Intelligence

Speaking about double-digit growth across iPhone, Mac, and Services, Cook said: "We were excited to introduce a beautiful new software design that extends across all of our platforms, and we announced even more great Apple Intelligence features."

The company has fallen behind the likes of Microsoft and Google, who are spending around $85-100 billion annually on AI and data centers. Currently, Apple mostly relies on third-party data centers and has built a handful of AI tools in house, however users have been left facing disappointing launches and delays.

All of this could be about to change if Bloomberg reporting suggesting that Apple could buy Perplexity turns out to be true.

Speaking about data center spend on the earnings call, CFO Kevan Parekh explained: "It's not going to be exponential growth, but it is going to grow substantially."

However, Apple's cautious approach has proven valuable in terms of the company's environmental impact, where other companies have seen significant impacts from rapid data center expansion. In its 2024 Environmental Progress Report, the company highlighted its own proprietary server designs, which have boosted energy efficiency and reduced water consumption – they've also been powered by renewable energy since 2018.

You might also like

• We've listed the best AI tools and best AI writers
• Fancy an upgrade? These are the best Macs and MacBooks
• 'Apple must do this, Apple will do this' – Tim Cook rallies staff in face of AI delays

It’s pretty much common knowledge by this point that Siri is lacking compared to rival assistants and artificial intelligence (AI) tools like ChatGPT. But according to a new report, Apple might be about to inject it with some much-needed power that could potentially transform it into a true force in AI.

In the latest edition of his Power On newsletter, Bloomberg journalist Mark Gurman claims that Apple is essentially building its own ChatGPT rival that would bring chatbot capabilities to Siri.

The company has reportedly created its own 'Answers, Knowledge and Information' team to do this – and if it works, it could fix one of the biggest gripes I have with Siri. Too often, I ask Siri a question, only for it to bring up a search engine results page and cheerily respond with “Here’s what I found on the web.”

Other times, it’ll suggest handing over to ChatGPT instead of answering the question itself. For example, just try asking Siri a relatively simple question – “what is the most common type of tree in North America?”, for instance – and it’ll ask if you want to enlist ChatGPT for this task.

Both outcomes highlight how incapable Siri is compared to AI tools like ChatGPT. It’s like Siri is hoisting the white flag. “I don’t know what the answer is, but maybe you can find it yourself?” seems to be the message it’s conveying.

(Image credit: Shutterstock)

And there’s another aspect to this. Whereas ChatGPT can actually find the answer for me, Siri’s response is the equivalent of those sarcastic “let me Google that for you” web pages that rudely hint at your own stupidity for not simply searching for the answer yourself. That’s never a good look for a personal assistant like Siri.

But the bigger problem is that handing off to Google or ChatGPT doesn’t actually solve the problem I wanted a solution for in the first place, and it misses the entire reason why people ask Siri questions in the first place.

People ask Siri because they want answers, not because they want to do the heavy lifting themselves. If the outcome I’m looking for is a Google search results page, I could have opened Safari and done that myself. I wanted an answer, not more hunting around.

And that’s the key difference between tools like ChatGPT and Siri right now: ChatGPT provides answers, while Siri does not. Even when Siri ropes ChatGPT into the situation, it’s an extra step compared to Siri just giving you the result you wanted. Whatever the question, Siri is rarely the answer.

A shot in the arm

(Image credit: Apple)

Hopefully, Apple adding these “answer” capabilities to Siri could be the shot in the arm its virtual assistant desperately needs. That prospect is certainly more exciting to me than some of the other delayed Siri features, like its ability to work within apps or understand your personal context.

While those features are interesting and might grab the headlines, I don’t see myself using them very often. They’re more complex and specialized, whereas asking Siri a simple question is a much more common occurrence for me and, I suspect, the majority of Apple users.

Still, questions remain over how Apple will transform Siri into a proper answer engine. How will the company train Siri to be able to answer more questions? Will Siri simply search online for results, or will it be able to rely on stored knowledge? How will Siri be trained in a way that aligns with Apple’s much-vaunted privacy principles? And will it take away traffic from third-party websites, withering the web in the way ChatGPT and Google’s own AI overviews are threatening to do?

And then there’s the question of when Siri’s chatbot update will arrive. Apple is reportedly targeting a spring 2026 launch date for its next-generation Siri features, but that hasn’t previously included any mention of a chatbot, which could take a long time to develop. With AI developments moving at breakneck speed, will a Siri-based chatbot be able to keep up with its rivals?

Those questions will be answered in due course, but in the meantime, I’m looking forward to a more capable (and, frankly, less useless) virtual assistant. If Apple can offer these features in a way that respects your privacy and doesn’t rip off the world’s online content, then I won’t mind if Siri is a little less capable than ChatGPT. As long as it’s more capable than it is right now.

After all, the whole point of using an AI helper for search is to get actual answers, not a list of results that I could have searched for myself. If Apple can pull it off, this could be the upgrade Siri has been in dire need of for years.

You might also like

• Apple reportedly targets 'spring 2026' for launch of delayed AI Siri upgrade – but is that too late?
• 4 ChatGPT features I hope Apple adds to Siri
• Apple officially delays the AI-infused Siri and admits, ‘It’s going to take us longer than we thought’

• A South Korean YouTuber is folding and unfolding the Samsung Galaxy Z Fold 7 200,000 times in an extreme durability test
• The test is, at the time of writing, still being livestreamed on YouTube
• The phone has survived more than 130,000 near-consecutive folds

A South Korean tech YouTuber has taken on the challenge of unfolding and folding the Samsung Galaxy Z Fold 7 200,000 times, putting Samsung’s latest flagship foldable through an extreme durability test.

YouTube channel Tech It is, at the time of writing, still streaming the experimental test, marking interesting developments as they go.

At 6,000 fold-unfold cycles, the phone suffered a reboot error, possibly overwhelmed by the constant switching between its folded and unfolded states.

At 46,000 cycles, the hinge began to creak – not encouraging at such a low number of folds, but keep in mind that this test is the furthest possible thing from everyday use.

At 75,000 cycles, text overlaid on the screen says an “unidentified liquid” began to leak from the hinge. Again, this is intriguing, but shouldn’t be cause for alarm unless you also plan to put your Galaxy Z Fold 7 through several thousand folds consecutively.

At the time of writing, the livestream has been running for nine hours, with just over 137,000 fold-unfold cycles registered. That suggests an average pace of around 4 cycles per second, though the current pace seems to have slowed to around 2.5 cycles per second.

At their current pace, and at the time of publishing, Tech It will complete their mission in around seven hours.

Should you pay attention to durability tests?

The Galaxy Z Fold 7 is more durable than its predecessor on paper, but that's a hard claim to accurately test. (Image credit: Lance Ulanoff / Future)

As we previously reported, Samsung claims the Galaxy Z Fold 7 can withstand 500,000 folds – assuming you unfold your phone 50 times a day, that’s around 27 years of daily usage.

However, there’s no truly useful way to put this claim to the test – it’s not practical, or sensible, for the everyday user to count and log each time they unfold their phone.

Here at TechRadar, we don’t test for durability – though phone makers are always improving the resilience and longevity of their devices, there’s simply no objective way to test most durability claims.

Even quantifiable limits, like the 500,000 fold promise made by Samsung, can rely on a number of factors like temperature, frequency of use, or manufacturing inconsistencies.

Still, there is some value in tests like this – in the latest infamous bend test from YouTuber Jerry Rig Everything, the Samsung Galaxy Z Fold 7 refused to break in half – that's about as blunt a measurement as you could possibly take, but still noteworthy for a hinged device.

So while these tests are fun and eye-catching, be sure to check out our guides to the best folding phones and the best Samsung phones to find out how these handsets perform in real-world usage.

You might also like

• This rumored Samsung Galaxy S26 upgrade will finally make contactless payments as easy as they are on the iPhone
• We may now have a full specs sheet for the upcoming Samsung Galaxy S25 FE
• The Galaxy Z Fold 7 costs too much – but this could actually be a good thing for folding phones

From new episodes of The Summer I Turned Pretty season 3 to the premiere of Amazon Originals like The Pickup (pictured above), Prime Video has a stacked list of new titles being added throughout the month.

August's lineup is slightly smaller then everything new on Prime Video in July, but there are notably more Amazon Originals to look out for, including two sporting docuseries Built in Birmingham: Brady & the Blues and Taurasi, the spy thriller Butterfly, the animated series Sausage Party: Foodtopia and the romantic drama The Map That Leads to You.

As usual, these titles will be joined by a bumper crop of licensed movies from the Hollywood vaults, including 80s classics such as Raging Bull and Sixteen Candles, and various live sporting matches from the New York Yankees, WNBA, and more. Read on for the full list of everything new on one of the best streaming services.

Everything new on Prime Video in August 2025

Arriving August 1

30 Rock seasons 1–7 (2006) – TV show

A Guy Thing (2003) – movie

Big Fat Liar (2002) – movie

Blow Out (1981) – movie

Built in Birmingham: Brady & the Blues (2025) – TV show

Cape Fear (1991) – movie

Conan the Barbarian (1982) – TV show

Conan the Destroyer (1984) – TV show

Death Becomes Her (1992) – movie

Death Race (2008) – movie

Death Race (Unrated) (2008) – movie

Duck, You Sucker! A Fistful of Dynamite (1972) – movie

Hazlo como hombre (2017) – movie

Hercules (2014) – movie

Howard the Duck (1986) – movie

King Solomon's Mines (1985) – movie

Lady Chatterley's Lover (2022) – movie

Lone Survivor (2013) – movie

Love Actually (2003) – movie

Maid in Manhattan (2002) – movie

Mermaids (1990) – movie

Miami Vice (2006) – movie

Mr. Mom (1983) – movie

Navy SEALS (1990) – movie

Only the Brave (2017) – movie

Out of Time (2003) – movie

Over the Top (1987) – movie

Overboard (2018) – movie

Pulp Fiction (1994) – movie

Raging Bull (1980) – movie

Ranchlands season 1 (2019) – TV show

Red Dragon (2002) – movie

Sixteen Candles (1984) – movie

Sleepover (2004) – movie

Something New (2006) – movie

Stigmata (1999) – movie

The Alamo (2004) – movie

The Battle of Britain (1969) – movie

The Return of the Living Dead (1985) – movie

The Strangers (2008) – movie

Transformers (2007) – movie

Transformers: Dark of the Moon (2011) – movie

Transformers: Revenge of the Fallen (2009) – movie

Uptown Girls (2003) – movie

Walking Tall (2004) – movie

Walking Tall: The Payback (2007) – movie

Wrath of Man (2021) – movie

Yours, Mine and Ours (1968) – movie

August 2

Kingdom of the Planet of the Apes (2024) – movie

August 6

My Dead Boyfriend (2016) – movie

The Pickup (2025) – movie

The Summer I turned Pretty season 3 episode 5 (2025) – TV show

August 7

Taurasi (2025) – TV show

August 11

Father Stu (2022) – movie

August 13

Butterfly season 1 (2025) – TV show

Run the Tide (2016) – movie

Sausage Party: Foodtopia season 2 (2025) – TV show

The Summer I Turned Pretty season 3 episode 6 (2025) – TV show

August 15

Betty la Fea: La Historia Continúa season 2 (2025) – TV show

The Siege at Thorn High (2025) – movie

August 16

Creed (2015) – movie

August 18

Homefront (2013) – movie

August 20

Pocket Listing (2016) – movie

The Map That Leads to You (2025) – movie

The Summer I Turned Pretty season 3 episode 7 (2025) – TV show

August 21

Thursday Night Football (2025) – TV show

August 22

The Intern (2015) – movie

Wolf Man (2025) – movie

August 25

Upload season 4 (2025) – TV show

August 27

The Summer I Turned Pretty season 3 episode 6 (2025) – TV show

The Terminal List: Dark Wolf (2025) – TV show

August 29

Last Breath (2025) – movie

August 30

Hannibal season 1–3 (2013) – TV show

You might also like

• Everything new on Netflix in August 2025
• Everything new on Paramount+ in August 2025
• Everything new on Hulu in August 2025

• Valve's SteamOS market share has grown by 0.32%, based on new Steam Hardware & Software survey
• Microsoft's Windows 11 market share fell by 0.44%, suggesting gamers are migrating to SteamOS
• These results come right before Microsoft's new 'full-screen experience' for the Xbox PC app

Microsoft's Windows 11 is on the verge of receiving significant updates leading into 2026, specifically for gamers with its new Xbox PC app 'full-screen experience'. However, it appears as though Valve's Linux-based operating system is gradually taking the lead as the better operating system for gaming.

As reported by TweakTown, Steam's July 2025 Hardware & Software survey results suggests gamers are switching to Valve's SteamOS from Microsoft's Windows 11, with a 0.32% growth in Linux market share – while the market share for Windows 11 fell by 0.44%.

TweakTown highlights that Linux's 2.89% operating system market share equates to around four million players out of Steam's 132 million users, so the growth is fairly significant. It may also be a huge indication of how popular handheld gaming has become since Valve introduced its Steam Deck back in 2022, if a majority of these results stem from handheld gamers.

It's quite clear that Valve's SteamOS still has a long way to go in terms of its user count compared to Windows 11, but the Linux operating system has arguably already surpassed Windows in terms of game performance, accessibility, and features.

It's also worth noting that both Valve and Microsoft have pledged commitments to improve their respective operating systems. The former plans to expand SteamOS's compatibility with other handhelds and desktop PCs, while Microsoft is bringing a 'full-screen experience' to handhelds, debuting on the ROG Xbox Ally.

Analysis: I doubt the new Xbox PC app 'full-screen experience' will check the SteamOS momentum

(Image credit: Future)

As I've stated previously, it's very hard for me to see how Microsoft's upcoming Xbox PC app experience will shift me away from SteamOS, because the latter has simply won me over (at least with handhelds).

I won't say Microsoft doesn't stand a chance, and I'm certainly going to be checking it out once it's available – but I think the stage is already set for Valve to take the lead when it comes to PC gaming.

The tools at my disposal while using Bazzite (a SteamOS clone) are great for UI customization, and perhaps most importantly, improved game performance with Decky Framegen and Decky Lossless Scaling, all via Decky Loader.

Sure, there are certainly equivalent tools on Windows 11, but navigation and tinkering on this operating system, especially with a handheld, is something I don't want to deal with anymore. That's why this 'full-screen experience' sounds promising, as it sounds like an equivalent to SteamOS's game mode.

Microsoft's new efforts to improve gaming on its operating system are very welcome and appreciated, and I think it will attract many gamers once it launches with the ROG Xbox Ally – but overall, I suspect it won't check SteamOS's momentum.

You might also like...

• I don't see how Microsoft's Xbox PC app will win me over – especially when I'm obsessed with SteamOS
• Forget the Blue Screen of Death – Windows is replacing it with an even more terrifying Black Screen of Death
• I'm absolutely sick of Microsoft's Windows 11 24H2 update, as it's now causing Bluetooth and webcam issues